As integrated circuit devices continue to shrink in size, the demand to integrate different functionalities on the same integrated circuit die also continues to grow. For example, portable wireless communication products have become high volume consumer devices. Some of these devices are now operating in the 1-2 GHz frequency range. There is, as a consequence, a demand to integrate RF front end circuits into high-yield silicon integrated circuit processes to allow a combination of analog, digital, and RF functions on the same integrated circuit die. Yet, considerable difficulty has been experienced in attempts to fabricate inductors having high quality factors (Q) in silicon technology for RF circuits which are used in communications.
Basic single transistor RF oscillator configurations such as Colpitts, Hartley, and Clapp configurations using FETs are known in hybrid electronics technology and are typically implemented using discrete components along with integrated circuit FETs. Examples are described in U.S. Pat. No. 5,138,285, the disclosure of which is incorporated by reference. In gallium arsenide monolithic microwave integrated circuits (MMICs), where all of the components are integrated on the same semiconductor substrate, advantages can be attained in that both the FETs and the inductors can be fabricated on the same semi-insulating substrate which results in minimal parasitic losses and high quality factor (Q) spiral inductors.
Voltage controlled oscillators or VCO's, have been reported in CMOS integrated circuit technology using spiral inductors, but, because of the difficulty in fabricating spiral inductors, the inductors have relatively small values, e.g. 3.2 nH, and large capacitors, e.g. 2.44 pF, are required to form a desired oscillator. With component values such as these, wafer real estate demands can be prohibitive. This can, as a result, cause inconsistencies in integration with other CMOS analog and digital circuits.
Attempts have been made to build high-Q inductors in silicon integrated circuit technology, but have yielded Q factors of only three to eight. This is undesirable in the context of RF circuit design at frequencies in the above-stated range, where required Q factors need to be typically in a range from between 5-20 for broad-band applications, and may have to be higher than 30 in narrow-band networks. Problems associated with the use of silicon technology in these scenarios, in part, result from the conductivity of silicon substrates which tends to induce losses. As frequencies approach the self resonant frequency, the inductance value decreases which is most undesirable. Losses in the conductive silicon substrates can be increased by the high dielectric constant of the insulators under the conductors and the relatively large values of stray capacitance coupling to the silicon substrate.
Some attempts have been made to provide oxide-encased, spiral-type inductors for silicon technology, with such encased inductors being disposed over a cavity which is etched into the silicon substrate. Others have attempted to provide higher-Q inductors in a five or six-level metal BiCMOS technology. The conductors in these instances are still encased in oxide but are far removed from the silicon substrate by virtue of a large number of insulator and metal levels. The number of these levels, however, is far in excess of the two to four levels commonly utilized in CMOS technology.
Other attempts have been made to provide higher-Q inductors through the use of long pad-to-pad wire bond techniques in BiCMOS technology in the design and fabrication of voltage controlled oscillators. Wire bond inductors have previously been used in a variety of applications as inductors and for impedance matching networks and, more recently, to create low impedance resonant connections from guard rings or bonding pads to ground planes. Having the conductor surrounded by air rather than an insulator serves to reduce losses from the conductive silicon substrates and yield high-Q values (11 to 15 at 1.8 GHz). However, having such long unsupported spans of wire, e.g. up to three millimeters in some cases, does not provide for good mechanical stability. In addition, there is no provision nor is there a possibility of passivation in these structures. Air bridge or wire bond inductors might have very good high-Q values, but their characteristics typically tend to be subject to change in the event of severe mechanical shocks or abrasion. In addition, such structures are not suitably protected from corrosive environments.
Against the backdrop of these attempts, there continues to remain a need in silicon integrated circuit technologies for high-Q inductors with rigid and fixed mechanical characteristics. Such inductors can be used in the design and implementation of oscillators, tuned amplifiers, and in optimizing broad band amplifiers. These cannot be achieved by standard integrated circuit techniques, nor air bridge and/or suspended metal conductors.
Accordingly, this invention arose out of concerns associated with providing improved oscillators and methods of forming the same. This invention also arose out of concerns associated with providing improved monolithic oscillators capable of functional performance at the above-stated frequencies.